Content : Graphics Processing Units (GPUs) have seen a fast-paced adoption in the context of High Performance Computing (HPC), with increasingly more applications benefiting from the massive parallelism of These architectures.While the performance promises or GPUs are impressive and tempting, making use of These processors requires a good understanding of Their hardware properties, the ability to work with a few parallel processing concepts, and basic knowledge in performance evaluation and analysis.
The goal of this course is to cover three areas thesis Such That participants are bootable to design and implement GPGPU applications of moderate difficulty and analyze Their Performance.Therefore, the course Consists of a 2h lecture for presenting the concepts and explaining the main challenges related to GPGPU programming, followed by a 2h hands-on practical session, where real applications will be Implemented and Analyzed.
- Duration:. 4 hours
- Prerequisites: C-programming. Basic concepts are useful parallelism.
- Course leader Ana Varbanescu (Informatics Institute)